Application of wavelet packet analysis and artificial neural network to concrete flaw in ultrasonic testing
ZHANG Li-xin~1 GAO Bai-feng~1 LI Hui-jian~1 (1.College of Architecture Engineering and Mechanics,Yanshan University,Qinhuangdao,Hebei 066004,China)
According to the nonstationarity of ultrasonic signals in ultrasonic testing,a method of flaw recognition based on the combination of wavelet packet analysis with ANN was presented in this paper.Here,wavelet packet analysis is used to extract char- acteristic values reflecting the flaw properties and back propagation neural networks are used to classify the characteristic values. A series specimen with hole was made to test the method above.Results show that by this method human effects on qualitative rec- ognition of flaws can be reduced,and high accuracy of flaw recognition can be obtained.
【CateGory Index】： TU755.7